Halfspace depth (HD), aka Tukey depth, is one of the most prevailing depth notions among all its competitors. To exactly compute the HD in is a challenging task nevertheless, due to its definition involving infinitely many directional projections. Existing algorithms to compute HD in (d ⩾ 2), more or less involve data projection to the directions perpendicular to hyperplanes (therefore involving polytopes or polyhedral cones) are either dimension d-, or sample size n-limited, or slow. Thanks to the work of Merkle and Bogićević and Merkle, algorithms for the fast computation of HD in ultra high dimension d for large n are proposed and addressed. They are sheer ball-based (or essentially point-wise distances) computations. The worst time complexity to approximately compute the depth m/n of a single point is O((d + m + log (n))n2) which seems to be one of the best known results so far among the competitors for extra large d and n. Unlike most previous algorithms, they do not require the underlying dataset to be in general position. As by-products of the algorithm, some depth regions (or depth level sets), and depth of sample points also could be obtained during the depth calculation of a given point.
Trimmed (and winsorized) standard deviations based on a scaled deviation are introduced and studied. The influence functions and limiting distributions are obtained. The performance of the estimators with respect to high breakdown scale estimators is evaluated and compared. Unlike other high breakdown estimators which perform poorly for light-tailed distribution and when points near the centre are contaminated, the resulting trimmed (and winsorized) standard deviations are much more efficient than their predecessors at light-tailed distributions by suitably choosing the cutting parameter and highly efficient for heavy-tailed and skewed distributions. At the same time, they share the best breakdown point robustness of the sample median absolute deviation for any common trimming thresholds. Compared with their predecessors, they can achieve the best efficiency when the contaminating points are presented from areas around the centre. Indeed, the scaled-deviation-trimmed (winsorized) standard deviations behave very well overall and, consequently, represent very favourable alternatives to existing scale estimators.